xs_url.h 6.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263
  1. /* copyright (c) 2022 - 2025 grunfink et al. / MIT license */
  2. #ifndef _XS_URL_H
  3. #define _XS_URL_H
  4. xs_str *xs_url_dec(const char *str);
  5. xs_str *xs_url_enc(const char *str);
  6. xs_dict *xs_url_vars(const char *str);
  7. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header);
  8. #ifdef XS_IMPLEMENTATION
  9. xs_str *xs_url_dec(const char *str)
  10. /* decodes an URL */
  11. {
  12. xs_str *s = xs_str_new(NULL);
  13. while (*str) {
  14. if (!xs_is_string(str))
  15. break;
  16. if (*str == '%') {
  17. unsigned int i;
  18. if (sscanf(str + 1, "%02x", &i) == 1) {
  19. unsigned char uc = i;
  20. if (!xs_is_string((char *)&uc))
  21. break;
  22. s = xs_append_m(s, (char *)&uc, 1);
  23. str += 2;
  24. }
  25. }
  26. else
  27. if (*str == '+')
  28. s = xs_append_m(s, " ", 1);
  29. else
  30. s = xs_append_m(s, str, 1);
  31. str++;
  32. }
  33. return s;
  34. }
  35. xs_str *xs_url_enc(const char *str)
  36. /* URL-encodes a string (RFC 3986) */
  37. {
  38. xs_str *s = xs_str_new(NULL);
  39. while (*str) {
  40. if (isalnum(*str) || strchr("-._~", *str)) {
  41. s = xs_append_m(s, str, 1);
  42. }
  43. else {
  44. char tmp[8];
  45. snprintf(tmp, sizeof(tmp), "%%%02X", (unsigned char)*str);
  46. s = xs_append_m(s, tmp, 3);
  47. }
  48. str++;
  49. }
  50. return s;
  51. }
  52. xs_dict *xs_url_vars(const char *str)
  53. /* parse url variables */
  54. {
  55. xs_dict *vars;
  56. vars = xs_dict_new();
  57. if (xs_is_string(str)) {
  58. /* split by arguments */
  59. xs *args = xs_split(str, "&");
  60. const xs_val *v;
  61. xs_list_foreach(args, v) {
  62. xs *dv = xs_url_dec(v);
  63. xs *kv = xs_split_n(dv, "=", 1);
  64. if (xs_list_len(kv) == 2) {
  65. const char *key = xs_list_get(kv, 0);
  66. const char *pv = xs_dict_get(vars, key);
  67. if (!xs_is_null(pv)) {
  68. /* there is a previous value: convert to a list and append */
  69. xs *vlist = NULL;
  70. if (xs_type(pv) == XSTYPE_LIST)
  71. vlist = xs_dup(pv);
  72. else {
  73. vlist = xs_list_new();
  74. vlist = xs_list_append(vlist, pv);
  75. }
  76. vlist = xs_list_append(vlist, xs_list_get(kv, 1));
  77. vars = xs_dict_set(vars, key, vlist);
  78. }
  79. else {
  80. /* ends with []? force to always be a list */
  81. if (xs_endswith(key, "[]")) {
  82. xs *vlist = xs_list_new();
  83. vlist = xs_list_append(vlist, xs_list_get(kv, 1));
  84. vars = xs_dict_append(vars, key, vlist);
  85. }
  86. else
  87. vars = xs_dict_append(vars, key, xs_list_get(kv, 1));
  88. }
  89. }
  90. }
  91. }
  92. return vars;
  93. }
  94. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header)
  95. /* parses a multipart/form-data payload */
  96. {
  97. xs *boundary = NULL;
  98. int offset = 0;
  99. int bsz;
  100. char *p;
  101. /* build the boundary string */
  102. {
  103. xs *l1 = xs_split(header, "=");
  104. if (xs_list_len(l1) != 2)
  105. return NULL;
  106. xs *t_boundary = xs_dup(xs_list_get(l1, 1));
  107. /* Tokodon sends the boundary header with double quotes surrounded */
  108. if (xs_between("\"", t_boundary, "\"") != 0)
  109. t_boundary = xs_strip_chars_i(t_boundary, "\"");
  110. boundary = xs_fmt("--%s", t_boundary);
  111. }
  112. bsz = strlen(boundary);
  113. xs_dict *p_vars = xs_dict_new();
  114. /* iterate searching the boundaries */
  115. while ((p = xs_memmem(payload + offset, p_size - offset, boundary, bsz)) != NULL) {
  116. xs *s1 = NULL;
  117. xs *l1 = NULL;
  118. const char *vn = NULL;
  119. const char *fn = NULL;
  120. const char *ct = NULL;
  121. char *q;
  122. int po, ps;
  123. /* final boundary? */
  124. p += bsz;
  125. if (p[0] == '-' && p[1] == '-')
  126. break;
  127. /* skip the \r\n */
  128. p += 2;
  129. /* Tokodon sends also a Content-Type headers,
  130. let's use it to determine the file type */
  131. do {
  132. if (p[0] == 13 && p[1] == 10)
  133. break;
  134. q = strchr(p, '\r');
  135. /* unexpected formatting, fail immediately */
  136. if (q == NULL)
  137. return p_vars;
  138. s1 = xs_realloc(NULL, q - p + 1);
  139. memcpy(s1, p, q - p);
  140. s1[q - p] = '\0';
  141. if (xs_startswith(s1, "Content-Disposition") || xs_startswith(s1, "content-disposition")) {
  142. /* split by " like a primitive man */
  143. l1 = xs_split(s1, "\"");
  144. /* get the variable name */
  145. vn = xs_list_get(l1, 1);
  146. /* is it an attached file? */
  147. if (xs_list_len(l1) >= 4 && strcmp(xs_list_get(l1, 2), "; filename=") == 0) {
  148. /* get the file name */
  149. fn = xs_list_get(l1, 3);
  150. }
  151. }
  152. else
  153. if (xs_startswith(s1, "Content-Type") || xs_startswith(s1, "content-type")) {
  154. l1 = xs_split(s1, ":");
  155. if (xs_list_len(l1) >= 2) {
  156. ct = xs_lstrip_chars_i(xs_dup(xs_list_get(l1, 1)), " ");
  157. }
  158. }
  159. p += (q - p);
  160. p += 2; // Skip /r/n
  161. } while (1);
  162. /* find the start of the part content */
  163. if ((p = xs_memmem(p, p_size - (p - payload), "\r\n", 2)) == NULL)
  164. break;
  165. p += 2; // Skip empty line
  166. /* find the next boundary */
  167. if ((q = xs_memmem(p, p_size - (p - payload), boundary, bsz)) == NULL)
  168. break;
  169. po = p - payload;
  170. ps = q - p - 2; /* - 2 because the final \r\n */
  171. /* is it a filename? */
  172. if (fn != NULL) {
  173. /* p_var value is a list */
  174. /* if filename has no extension and content-type is image, attach extension to the filename */
  175. if (strchr(fn, '.') == NULL && ct && xs_startswith(ct, "image/")) {
  176. char *ext = strchr(ct, '/');
  177. ext++;
  178. fn = xs_str_cat(xs_str_new(""), fn, ".", ext);
  179. }
  180. xs *l1 = xs_list_new();
  181. xs *vpo = xs_number_new(po);
  182. xs *vps = xs_number_new(ps);
  183. l1 = xs_list_append(l1, fn);
  184. l1 = xs_list_append(l1, vpo);
  185. l1 = xs_list_append(l1, vps);
  186. p_vars = xs_dict_append(p_vars, vn, l1);
  187. }
  188. else {
  189. /* regular variable; just copy */
  190. xs *vc = xs_realloc(NULL, ps + 1);
  191. memcpy(vc, payload + po, ps);
  192. vc[ps] = '\0';
  193. p_vars = xs_dict_append(p_vars, vn, vc);
  194. }
  195. /* move on */
  196. offset = q - payload;
  197. }
  198. return p_vars;
  199. }
  200. #endif /* XS_IMPLEMENTATION */
  201. #endif /* XS_URL_H */